No matter the size of the electrical grid, the rule of survival never changes: you must balance your power sources with your loads. For traditional commercial grids, this meant burning fuel or harnessing water to physically spin a generator and induce electricity. Today’s microgrids face that same balancing act, but with a twist. We must manage a complex mix of traditional spinning generators and static, silent sources like solar power.

For my load test, we cut the cord with our local utility—Lee County Electrical Cooperative (LCEC)—and spun up the N4FTD Power Grid. From the moment I flipped the transfer switch on Saturday afternoon until I conceded to LCEC around midnight on Sunday, I was officially on duty as a micro-utility operator. I pulled triple duty during those 34 hours: full-time caregiver for Sue, grid operator, and data collector feeding live telemetry to my AI companion, Gemini. This post is the deep dive into what that data revealed during our two-day After-Action Report (AAR).

Our Early Math & Testing Individual Components

I’ve been working on this N4FTD Power Grid project for a few months now. My earliest data consisted of anecdotal reports from my friend Carmine, our radio club’s president who happens to own Ecoflow gear. He has 10kWh; I have 8. I’m not sure if he runs an air conditioner with his system. After picking his brains for ideas I finally purchased my own hardware a few months ago. When my Ecoflow gear arrived, I conducted mini-tests to see how long I could run the fridge-freezer on my DP3 (4000Wh). Then, I plugged the depleted unit into a 120V house outlet to see how long it would take to fully recharge my Ecoflow batteries back up  to 100% SOC (state of charge). I recorded everything I learned in a Google Sheet.

Later, I decided to add solar and reasoned that a mix of solar + generator charging made sense for someone living in sunny Florida. I started out small with a single 400W panel just to see how it would do with my 4000W system. That’s when I learned that the Zoupw panels are highly efficient and can produce at or even above their rated output when in full sun. Then, I added two more 450W panels and continued the testing with excellent results. I have some readings showing outputs of almost 1kW using my two panels wired in series—111% of their rated output power. Those results went into my data spreadsheet as well. When I began using AI to “do the math” so I could see how all the component pieces were likely to work as a whole, we were quite excited about the possibilities. At least on paper, things were looking good.

Based on all this early data gathered and isolated testing of individual pieces of the system, Gemini and I came to the conclusion that we might be able to strike a balance between daytime charging and running on batteries alone during the overnight hours. If we could gather several daylight hours of solar charged power followed by an early evening generator charging session of up to 1.5 hours to top off the battery bank, we should be in great shape for the quiet and dark overnight hours. Before we bought the window A/C unit we added in its theoretical numbers, too. That was the first “educated guess” we made based solely on the manufacturer’s stats without any testing whatsoever to back up our math. It was then that I knew we needed a live “load test” to verify all of our combined data and the math we were relying on so much.

When “Best Case” and “Worst Case” Scenarios Meet

As you know from Part 1, our load test was scheduled for Saturday (noon) to Monday (noon), July 11-13, 2026. 48 Hours of charging-depleting-charging-depleting over 3 days. I made the executive decision to cut it short right before midnight Sunday. A look at the data reveals why I did that.

The Breakdown: Paper Assumptions vs. Heat Dome Reality

I gave Gemini a lot of data over those 34 hours. Now, it was time to ask for it all to be given back to me. This time, I asked the AI to compile and summarize all my inputs, draw conclusions, and show the cause-effect connections in chronological order. That way I’d be able to see how our inputs brought about the outcomes we experienced. Since this was live test data we were analyzing, we wanted our subjective experiences to be informed by the raw data and vice versa. This was more than an academic exercise. Our sweat equity was deeply embedded in every piece of data we analyzed.

Data Analysis

The Thermodynamic Deficit (The Math That Bended)

On paper, a 12,200 BTU portable A/C draws about 1,100W to 1,270W when the compressor is running.

• The Paper Assumption (Ideal Conditions, 50% Duty Cycle): The A/C cools the room to 80°F, the compressor shuts off, and it only runs ~50% of the hour (Duty Cycle = 0.5).
  • Average draw: 1,200W x 0.5 = 600W/hr.
  • Solar generation: 900W array produces ~600W–700W during peak daylight.
  • Math: Solar covers the A/C, excess trickles into the battery. Perfect balance.
• The Heat Dome Reality (100% Duty Cycle): Because the heat was radiating downward from the roof/attic through the vaulted ceiling, heat entered the Great Room faster than 12,200 BTUs could remove it.
  • Real-World Duty Cycle: 100% continuous run-time between 11:00 AM and 19:00 PM.
  • Real Draw: 1,270W nonstop.
  • Solar Degradation: Extreme thermal heat on solar panel cells drops their voltage efficiency by 15–20%. My 900W array maxed out at ~650W.
  • The Math Deficit: Input 650W – Load 1,270W = -620W/hr Deficit.
  • I was losing over 600 Watt-hours of battery capacity every single hour during peak sunlight!

That heat dome really threw us for a loop! At first, I was worried and frustrated. Then, I realized that it was a good thing. If we could make it work with a heat dome, surely a post-hurricane grid outage would be easier. I recalled post-Ian temps that sometimes allowed me to sleep quite well in my camper van with the windows opened and a fan running. No A/C needed.

The Recovery Drag (The Generator Reality Check):

• The Paper Expectation: Planning for a maximum of 60 to 90 minutes of generator runtime to top off the DP3 stack to 90–100% SOC.
• The Heat Dome Reality: On Day 2, it required 6 hours and 45 minutes of continuous generator runtime (from 14:15 to 21:00) just to crawl the bank from 5% back up to 85%.
• Why: Because the generator wasn’t just charging the batteries in a vacuum—it was forced to split its 2.88 kW DC output simultaneously carrying the active 1,270W A/C load while slowly trickling the remaining energy into heat-stressed battery cells.

This, too, worried me. As I suspected, the 30 lb tank felt much lighter the day after the test was cut short. I had counted on 90 pounds of propane to last me 2 weeks of 1-1.5 hour cycles. At the rate we burned it during this brief test we’d be lucky to get 5-6 days!

The Day vs. Night Trap

Dynamic	Daytime (The Solar Illusion)	Nighttime (The Thermodynamic Trap)
Primary Source	Solar Array + Battery Drain	8 kWh Battery Stack Alone (No Solar)
Ambient Air	96°F – 102°F	80°F – 85°F (No relief)
Radiant Heat	Sun actively heating roof	Walls/Ceilings releasing stored heat into the room
Compressor Action	Locked ON (1,270W continuous)	Cycling frequently, fighting radiant bleed
System Trajectory	Slow drain (-620W/hr deficit)	Fast drain (~750W/hr continuous draw)

The Sleeping Cliff: Starting the night at 85% capacity (6,800 Wh available), a continuous 750W draw drains the battery bank at ~9.3% per hour.

That leaves zero margin for error, zero battery floor buffer, and guarantees a dead system before sunrise.

Before I even had a chance to analyze the data I predicted that this wasn’t going to work. As it turned out, I was right.

The Parasitic Load & “Thermal Rebound”

When I turned the A/C compressor off for just 59 minutes at 23:02 to test baseline parasitic loads (fans, lights, Starlink, network gear drawing ~342W):

• Temperature Spike: The Great Room jumped +3.8°F (from 80.8°F up to 84.6°F) in under an hour.
• The Insight: Turning off the A/C to save battery backfires instantly. The room heats up so fast that when you turn the A/C back on, the compressor has to work twice as hard at maximum wattage for hours just to recover that lost ground.

After seeing that +3.8 degree jump I tried to sleep, but couldn’t. It wasn’t so much about the heat. It was my worry that Sue was also having to endure the heat. Add to that my frustration that after all our work, we were so far off-base. I realized that we would be completely without power before 6:00 AM without some sort of intervention. In a real post-hurricane neighborhood where everyone is struggling to “make do” while running their generators throughout the night, I would have simply fired up the generator. If you can’t beat ‘em with batteries, join ‘em with a propane generator! That’s why I bought it in the first place, “just in case”. This was only a test, though. Nobody else was running a generator at 11:30 PM in the middle of the night. All of those factors combining in my head while I lay there on the couch, sweating, was what finally led me to flip that big switch. I transferred back from my micro-grid to the LCEC commercial grid within minutes of all these factors converging in my mind. As I said in my first installment of this series, within a half-hour of turning on the “big air conditioner” both Sue and I were fast asleep. The AAR could wait until tomorrow; I needed a break from the action.

The Quadruple Role Challenge

Now, add in one more ingredient. I’ve spent over a year preparing myself, my family, and my CERT team for a post-hurricane, grid down scenario where we use a 3-Tiered Communications Plan to respond to our neighbors’ needs. As the comms leader, I have to lead the communications portion of that response. After the load test was finished and my wife and I began our AAR, I had a hard time imagining myself adding in a fourth role much less sustaining it for up to two weeks while running my own micro-grid and caring for Sue like I do every day in air conditioning with commercial power running quietly in the background. There’s no way I was going to sit in my super-hot shed while operating my radios and Starlink, then run out to start the generator or back into the house to fix lunch for my wife and me. No way! How is this ever going to work?

Early Ideas & A Possible Solution

My early ideas included a trip back to Ecoflow for another 4 kWh of batteries. That way, I’d have 2K more than my friend and mentor, Carmine! Or, maybe we could just skip the A/C and buy a few more fans instead. That would work. I knew it for a fact—actually, many facts, gathered along with all the other data during and prior to this load test.

It was during Day 2 of our two-day AAR that Sue actually gave me the seeds of an idea that just might work. I won’t spoil it this time. Let’s just say that when you’re in the wrong end zone and have over 100 yards to go for a score, sometimes you have to punt. Or, to mix metaphors, you might need to pivot. Let’s stick with The Pivot: How the N4FTD Power Grid project gets juiced! Come back next time to see how we hope to work out this massive power deficit. Hint: we’re not moving to Alaska!